Why attribution broke

Ad attribution in 2026 is less accurate than it was in 2020. iOS 14, cookie deprecation, and privacy regulation broke the old model. Advertisers who didn't adapt are operating blind. Those who did have new tools and frameworks.

What broke

iOS 14.5 (2021)

App Tracking Transparency gave users a choice to opt out of tracking. Most chose opt out. Pixel data from iOS users became incomplete.

Third-party cookies (phased out)

Chrome finally deprecated third-party cookies in 2024-2025. Cross-site tracking that powered retargeting died.

Privacy regulation

GDPR, CCPA, and successor laws make tracking more restricted and disclosure-heavy.

Safari and Firefox

Always ahead of Chrome on privacy. Strict third-party cookie blocking for years.

The consequences

• Platform-reported conversions undercount real conversions by 20-50%
• Multi-channel attribution is noisy
• Retargeting audiences shrank dramatically
• Lookalike audiences built on partial data are less precise
• Incrementality harder to verify

The adaptations

Server-side tracking (CAPI)

Send events server-to-server instead of relying on pixel. Meta CAPI, Google Enhanced Conversions.

First-party data

Build and use your own data (email lists, CRM, user behavior on your site) rather than third-party tracking.

Blended metrics

MER, ROAS overall, not per-channel. Accept that platform-specific numbers are directional.

Incrementality testing

Holdout tests, turn a channel off and measure revenue change. Truer than platform reports.

Marketing Mix Modeling

Statistical modeling of total spend vs total revenue. Doesn't need user-level tracking.

What it means for day-to-day

• Stop chasing exact platform-reported CAC; use directional signal
• Look at total blended CAC as the truth
• Invest in first-party data infrastructure
• Creative becomes more important (targeting is less precise)
• Build relationships with customers for data-sharing consent

What to do with this

• Stop treating platform-reported CAC as truth, it over-claims in most cases, use it for relative comparison only
• Shift to blended MER (total revenue / total spend) as your primary North Star, it's noisy but ground-truth
• Invest in first-party data (email list, CRM integration, post-purchase "how did you hear about us?"), it's the only data you truly own
• Run quarterly incrementality tests, pause a channel for 2 weeks and measure actual revenue drop, that's your real incremental value
• Accept that creative and offer now do more of the work than targeting, shift budget allocation to match